Describe, Don't Program

The Old Way

For decades, automation meant programming.

You had to think like a machine. Break tasks into atomic steps. Handle every possible exception. Anticipate every edge case. Build decision trees that covered every branch.

If you didn't specify something, it didn't happen. If you didn't handle an error, the whole thing crashed. The machine did exactly what you told it — nothing more, nothing less.

This required a specific kind of thinking. Technical thinking. And most business owners either had to learn it or hire someone who had.

The automation was only as good as your ability to think mechanically.

The New Way

But here's the thing: machines have changed.

AI agents don't need step-by-step instructions. They need to understand what you want. Give them an outcome, and they figure out the path. Give them a problem, and they work toward a solution.

This means the skill that matters has shifted.

It's no longer "can you program a solution?" It's "can you describe what you want?"

And describing what you want is something you already know how to do. You do it every time you explain a task to a colleague. Every time you write a brief. Every time you tell someone "here's what I need."

The technical barrier is gone. What remains is clarity.

The Art of Clear Description

What does good delegation to AI look like? The same thing good delegation to humans looks like:

The pattern is simple: treat your AI orchestra like a smart team member who needs to understand the mission, not a machine that needs to be programmed.

Your First Description

Here's a framework for thinking about any task you want to delegate:

1. 1.
   What's the outcome?What does "done" look like? What will exist that doesn't exist now?
2. 2.
   What context matters?What would a smart person need to know to do this well?
3. 3.
   What does good look like?How will you know if the result is right?

That's it. Answer those three questions, and you've written a workflow description.

That's a workflow. You just wrote it. No programming required.

What Happens Next

Each description teaches you to describe better.

Your first workflows will be rough. You'll realize you weren't clear about something. The output won't quite match what you had in your head. That's fine — that's learning.

But with each iteration, you get better at the art of clear description. You learn what context matters. You learn how to specify "good." You develop intuition for how to express what you want.

This is a skill that compounds. The better you get at describing, the more you can delegate. The more you delegate, the more practice you get. The more practice, the better you get.

And here's the beautiful part: this skill transfers. Getting better at describing what you want to AI makes you better at describing what you want to humans. It makes you better at thinking about work itself.

The Only Skill You Need

Programming required learning a new language — the language of machines.

Delegation requires something different — clarity about what you actually want.

You already know how to delegate — you do it every day with colleagues. But humans fill in gaps. They ask clarifying questions. They make reasonable assumptions. This forgiving environment means we rarely have to be perfectly clear.

AI raises the bar. Not because it's inflexible, but because it takes your words seriously. It does what you describe, not what you vaguely intended. Which means you have to mean what you say.

This is actually a gift. Because when you learn to describe what you really want, you often discover you weren't clear about it yourself. The act of describing forces clarity. And clarity is valuable everywhere.

So here's the invitation: stop trying to learn automation. Start learning to describe what you want.

The AI will handle the rest.